SqueezeNet: AlexNet-level accuracy with 50x fewer parameters and <0.5MB model size

24 Feb 2016Forrest N. IandolaSong HanMatthew W. MoskewiczKhalid AshrafWilliam J. DallyKurt Keutzer

Recent research on deep neural networks has focused primarily on improving accuracy. For a given accuracy level, it is typically possible to identify multiple DNN architectures that achieve that accuracy level... (read more)

PDF Abstract

Results from the Paper


TASK DATASET MODEL METRIC NAME METRIC VALUE GLOBAL RANK RESULT BENCHMARK
Image Classification ImageNet SqueezeNet Top 1 Accuracy 57.5% # 151
Top 5 Accuracy 80.3% # 109

Methods used in the Paper